Closed loop control systems are used in a variety of processes, including, by way of example, heating, venting and air conditioning (“HVAC”) applications. The purpose of such control systems is to control a process variable value x so that it is substantially equal to a setpoint w. If the setpoint w changes, then the control system endeavors to change the process variable x in response thereto.
By way of example, consider a setpoint TW identifying a desired temperature. The control system may control the operation of a heating vent or the flow of hot water to a heating coil in order to control the actual temperature represented as a process variable TX. If the actual temperature TX differs from the setpoint TW, then an error signal e is provided to a controller within the control system. The controller then acts upon the error signal e to determine a control signal to generate. The controller, as is known in the art, may employ a transfer function to generate the control signal based on the error signal. The transfer function may incorporate integration, derivation and proportional scaling of the error signal in an effort to create a control system that provides a balanced control of the actual temperature TX.
More specifically, because a control system cannot change the process variable instantaneously as a function of the control signal output, controllers employ sophisticated transfer functions to avoid excessive oscillations in response to changes in setpoint. For example, consider a situation in which setpoint temperature in a room of a building TW is 20° C. Assume also that the sensed temperature TX is 20° C. in steady state. If the setpoint temperature TW is changed from 20° C. to 21.3° C., then the control system may control a process, for example, the flow of hot water through a heating coil, in order to raise the temperature TX. The flow of hot water through the heating coil causes the sensed temperature TX to slowly rise. Once the ambient or sensed temperature TX reaches 21.3° C., the control system may decrease or stop the flow of hot water to the register. However, the hot water within the register will continue to heat because it cannot be cooled instantaneously. Thus, the temperature TX may exceed the set point. Because the temperature process variable exceeds the set point, the control system may turn off the flow of hot water completely. After some time, the temperature would cool to the desired 21.3° C. However, the temperature would continue to fall below 21.3° C. The control system may again cause hot water to flow to bring the temperature back to 21.3° C. but it takes some time for the heating coil to warm up and generate heat. The above described temperature oscillations may continue indefinitely if the control system is not properly tuned. Accordingly, the transfer function is typically chosen such that it reduces or eliminates the possibility of excessive oscillations in the control system.
As discussed above, the transfer functions of closed loop control systems employ known techniques to manipulation of the error signal before calculating the control output. For example, many control systems employ a proportional calculation in which only a fraction of every error signal is incorporated into the calculation of the control output. Proportional control thus tempers or reduces the effect any instantaneous error signal value will have on the output of the system, thereby reducing the potential for large oscillations. One popular form of controller, a PID controller, employs proportional, integrated and differentiated aspects of the error signal to formulate the control output. In a PID controller, the error signal is provided to a proportional circuit, a differentiating circuit, and an integrating circuit. The outputs of the circuits are combined to help generate the control output. The use of such differentiated and integrated aspects of the error signal further improves the response of the control system.
In any event, one issue that arises in control systems is their behavior when the error signal is very close to zero. More specifically, because of many factors, it is difficult to achieve absolutely zero error in control systems, particularly in large control systems such as HVAC control systems. These factors include noise and/or non-linearities generated by the mechanical equipment, the external environment, and other sources. The noise and nonlinearities introduce non-zero elements into the error signal, even though the nominal (noise free) error signal is zero. These non-zero elements can, without remediation, cause the control system to unnecessarily change its output signal.
Changing a control output typically causes actuation of a mechanical device, for example, movement of a heating vent, opening or closing of a valve, or change in fan speed. Thus, for example, a control output may frequently cause a heating vent to open and close in attempts to achieve zero error. Unnecessary actuation of mechanical devices typically shortens their life cycle. Accordingly, the difficulty in achieving zero error in control systems such as HVAC systems undesirably results in shorter life cycles for elements of the system.
In order to reduce the effects of noise and non-linearities, many control systems employ dead zone or dead zone operation. Dead zone operation typically involves a non-linear dead zone filter that generates an output value of zero if the error signal is within a predetermined range of zero. As a result of dead zone operation, control systems do not have to achieve zero error in order to avoid excess actuation of controlled devices. Thus, noise signals that could otherwise trigger actuation of a controlled device are filtered out by the dead zone filter. An example of a dead zone filter is discussed in U.S. Pat. No. 5,768,121.
The dead zone filtering of error signals in controllers has gained widespread acceptance. However, an issue with prior art dead zone operation is that the dead zone filter did not distinguish between steady state operation of the control system and transitional operation of the control system. In particular, when the control system has achieved a steady state near-zero error, then the dead zone operation provides the advantage of reducing or eliminating unnecessary operation of actuators due to noise on the error signal. However, when the control system is oscillating in an attempt to settle into steady state operation, the error signal may pass through the dead zone transitionally. In such a case, the dead zone operation does not provide any significant advantage, and indeed, introduces non-linearity into the controller which makes it difficult for the controller to settle.
Accordingly, there is an additional need for a controller that provides dead zone operation to reduce noise at steady state, but which does not introduce non-linearities into the controller when the controller is not at steady state.